SpectralLab
A web-based frequency-domain image analysis tool for visualization, filtering, and reconstruction across multiple transform chains (FFT/DFT, DCT, and DWT). Designed for fast research prototyping and teaching demos.
This project was produced with an LLM-assisted vibe coding workflow: rapid prototyping → iterative refactoring → polish.
Tooling: [ChatGPT GPT-5.2 Thinking / Codex CLI / VSCode ]
Links
• Live Demo: GitHub Pages
• Code: GitHub Repository
Highlights
• Multi-image workspace and transform switching
• Spectrum / coefficient visualization (log + normalization, FFT shift)
• Filters: ideal & Gaussian (LP/HP/BP/BS)
• Real-time reconstruction and side-by-side comparison
• Energy-based statistics panel
Ophthalmic Disease Diagnosis Platform
An ophthalmology-focused pipeline covering enhancement → segmentation → classification → Q&A. Centered on frequency-domain enhancement and FaRMamba segmentation, it emphasizes detail restoration and structured outputs for screening and research evaluation.
Awards
• China Robot and Artificial Intelligence Competition (CRAIC) 2025 (National Second Prize)
• Service Outsourcing Innovation & Entrepreneurship Competition (Eastern Division, Third Prize)
3D Medical Image Interactive Analysis Platform
An interactive platform for medical imaging that integrates 3D visualization, segmentation, and result exporting. It supports multi-modal data and basic measurement/annotation. A lightweight VLM module enables image-grounded Q&A and report drafting for teaching, research, and clinical assistance.
Awards
• Chinese Collegiate Computing Competition 2024 (National Second Prize)
• China Robot and Artificial Intelligence Competition (CRAIC) 2024 (National First Prize)